If you drew a black rectangle over text in a PDF, the text is still in the file. Select across the box, copy, paste — there it is. This single misunderstanding has exposed sealed testimony, confidential settlement terms, and personal data in public filings, repeatedly, for two decades. Real redaction means the content is removed from the file, and you can do it for free, in your browser, with the Redact PDF tool — no upload, no account.
Why the black box fails
A PDF is not a picture; it's a program that draws a page. Text lives in content streams as text objects. When you add a black rectangle in an annotation tool, you append another drawing instruction on top. The viewer renders text, then rectangle — and your eyes see black. The file, however, still contains every character.
Anything that reads the file rather than rendering it — text selection, search, copy-paste, screen readers, text extraction scripts — walks straight past the rectangle. That's why failed redactions are usually discovered within hours of a document becoming public: the recovery technique is selecting text.
Every cosmetic approach fails the same way:
| Method | Why it fails |
|---|---|
| Black rectangle / shape | Text object survives underneath |
| Black highlighting | Highlight is an annotation over intact text |
| White or transparent text color | Color changes rendering, not content |
| Cropping the page | Cropped areas are hidden, not deleted |
| Pasting an image over text | Image is a new layer; text remains |
The two-minute self-audit
Before any sensitive PDF leaves your hands, test it like an adversary would:
- Select across every redaction. Click before the box, Shift-click after it, copy, and paste into a plain-text editor.
- Search the document. Ctrl+F for names, numbers, and phrases you redacted. Try partial strings too.
- Select all, copy, paste. Ctrl+A on each redacted page, paste into a text editor, and read what comes out.
- Check the metadata. Author names, software identifiers, and revision data live outside the visible page entirely. The Metadata Viewer shows you everything riding along in the file and strips it.
If any test surfaces redacted content, the document was never redacted — it was decorated.
Redacting for real, in your browser
The Redact PDF tool does destructive redaction:
- Drop in your PDF — it loads into browser memory; nothing is uploaded anywhere.
- Draw boxes over everything that must go.
- Apply. The tool destroys the boxed regions and rebuilds each affected page as a flattened image. The output file contains no trace of the redacted content: no text object, no hidden layer, nothing to select or search.
- Run the self-audit above on the output. It should pass every test — that's the point.
The honest trade-off: rebuilt pages are images, so their text is no longer selectable for anyone, including legitimate readers. For documents where redaction matters, that trade is correct by definition — and you can keep an unredacted original under proper controls.
One more thing: don't upload what you're trying to protect
The standard online redaction workflow asks you to upload an unredacted document — the most sensitive version that will ever exist — to a stranger's server, and trust that it gets deleted. That's backwards. Whatever you think of any site's deletion policy, an unauditable copy of your secrets existed on someone else's machine.
PDF Cloak's tools run entirely client-side: open-source engines (pdf-lib, PDF.js) executing in your browser. Watch the network tab while you redact — zero requests. Turn off Wi-Fi — it still works. For redaction, local isn't a preference. It's the job.